🤖 Meta ships Muse AI agent at $20/month 🤖
Meta's first personal agent books travel, sends email and shops on your behalf.

Meta unveiled Muse, pitched as the world's first personal AI agent, powered by Muse Spark and running on cloud infrastructure. It automates everyday tasks — booking travel, sending emails, managing Spotify playlists, shopping — with integrations for Gmail, Ticketmaster and OpenTable and security handled by Muse Secure VM plus a “Sentinel” agent overseeing every action. The US rollout starts now across iOS, Android, WhatsApp and muse.ai, with subscription tiers at $20 or $100 per month.

📱 Apple ships iPhone Duo foldable at $1,999 📱
Apple's biggest rollout in years embeds AI across every new device under CEO John Ternus.

Apple held one of its most consequential keynotes in years — the first under new CEO John Ternus — with AI woven into every product announced. The star is the iPhone Duo, Apple's first foldable at $1,999, built around an AI-designed hinge, alongside an iPhone 18 Pro with a 48MP variable-aperture camera and AI-powered “Reference Image” pixel authentication. AirPods 5 add real-time Live Translation and the Apple Watch Series 12 introduces “Live Rewind”, transcribing the last 15 seconds of any conversation. Siri AI also launched in beta with iOS 27, bringing true conversational capabilities, onscreen awareness and cross-app personal context.

🛡️ Anthropic's 154-page report exposes global Claude misuse 🛡️
Anthropic's report reveals alarming misuse patterns challenging current AI regulation.

Anthropic released its first major threat intelligence report, 154 pages detailing misuse it disrupted between December 2025 and August 2026. The cases are stark: gain-of-function research on the chikungunya virus, Claude Code used in Yemen to build rocket guidance software, and a surveillance system designed to monitor 25 million phone lines in Mali. The report also names seven Chinese AI labs — including DeepSeek, Alibaba and Xiaomi — that used thousands of fake accounts to distill Claude's capabilities, some reselling its outputs as proprietary models. Two macro trends stand out: AI-enabled cyberattacks are growing more autonomous, and sophisticated attacks no longer require sophisticated attackers.

INTERESTING TO KNOW

📈 Anthropic models 32% US GDP boost by 2030 📈

Anthropic published an interactive simulator of how AI could reshape the US economy by 2030. Its most extreme scenario puts US GDP 32.4% above baseline with annual growth at 15.4%, but unemployment at 11.9% and knowledge workers hit hardest — 17.9% unemployment and wages down 11.5%. Labour's share of national income falls from 60% to 45.2% as capital income surges roughly 81%, with most effects staying muted until 2028.

🧮 OpenAI claims AI cracked Navier-Stokes $1M problem 🧮

OpenAI announced that an internal system produced a proof resolving the Navier-Stokes existence and smoothness problem, one of the seven Millennium Prize Problems and worth a $1 million bounty, by running roughly 10,000 agents simultaneously for 88 hours. NYU's Tristan Buckmaster and Anthropic's Levent Alpöge allege the proof uses a method they had been developing and uploaded to Codex. The full proof remains unpublished pending independent verification.